Multilingual phone models for vocabulary-independent speech recognition tasks
Speech Communication
Language-independent and language-adaptive acoustic modeling for speech recognition
Speech Communication
Language Modeling for Machine Translation
Language Modeling for Machine Translation
Hi-index | 0.00 |
In car infotainment systems commands and other words in the user's main language must be recognized with maximum accuracy, but it should be possible to use foreign names as they frequently occur in music titles or city names. Previous approaches did not address the constraint of conserving the main language performance when they extended their systems to cover multilingual input.In this paper we present an approach for speech recognition of multiple languages with constrained resources on embedded devices. Speech recognizers on such systems are typically to-date semi-continuous speech recognizers, which are based on vector quantization.We provide evidence that common vector quantization algorithms are not optimal for such systems when they have to cope with input from multiple languages. Our new method combines information from multiple languages and creates a new codebook that can be used for efficient vector quantization in multilingual scenarios. Experiments show significant improved speech recognition results.